Research map / 3D vision / Depth and stereo
Connecting the Dots: Learning Representations for Active Monocular Depth Estimation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use a fully convolutional network to directly regress disparity rather than absolute pattern position (ablation) | Active monocular depth estimation · Synthetic data | o(1): 5.08 -> 3.00 | Helped |
| Use shared fully convolutional prediction instead of independent per-row regression (ablation) | Active monocular depth estimation · Synthetic data | o(1): 4.35 -> 3.00 | Helped |
| Include the active projected pattern rather than predict disparity from ambient imagery alone (ablation) | Active monocular depth estimation · Synthetic data | o(1): 81.41 -> 3.00 | Helped |
| Add the disparity loss conditioned on ambient edges (ablation) | Photometric-loss-trained disparity network · Synthetic data | o(1): 6.00 -> 4.23 | Helped |
| Add geometric consistency loss across views (ablation) | Photometric + disparity loss model · Synthetic data | o(1): 4.23 -> 3.88 | Helped |
| Use all three losses instead of photometric supervision alone (ablation) | Active monocular depth estimation · Synthetic data | o(1): 6.00 -> 3.88 | Helped |
| Use the proposed self-supervised method rather than the strongest reported baseline | Active monocular depth estimation · Synthetic data | o(1): 7.20 -> 3.88 | Helped |
| Use the proposed method rather than HyperDepth | Active monocular depth estimation · Real data | h. mean: 6.575 -> 4.898 | Helped |
| Use the proposed method rather than HyperDepth | Active monocular depth estimation · Real data | acc: 8.759 -> 11.042 | Hurt |
Closest papers by meaning
- UltraStereo: Efficient Learning-Based Matching for Active Stereo Systems (CVPR 2017 · similarity 0.91)
- HyperDepth: Learning Depth From Structured Light Without Matching (CVPR 2016 · similarity 0.90)
- P3Depth: Monocular Depth Estimation With a Piecewise Planarity Prior (CVPR 2022 · similarity 0.90)
- Depth Pro: Sharp Monocular Metric Depth in Less Than a Second (ICLR 2025 · similarity 0.88)
- A Simple yet Universal Framework for Depth Completion (NeurIPS 2024 · similarity 0.88)
- Mind The Edge: Refining Depth Edges in Sparsely-Supervised Monocular Depth Estimation (CVPR 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.